@ Keith (or anyone who has mastered E@H)

. . I tried to change E@H to do CPU work as well as GPU. To get work on CPU I changed the resource share from 0 to 1 but now I am flooded with GPU tasks (hundreds of them) despite have work fetch set to 0.1 & 0 and trying 0.1 & 0.1. It just keeps replacing completed tasks so I am way overloaded ( between 2 and 3 days worth of work by my calculations). But still NO CPU tasks, it keeps saying I need to set preferences to allow them. I have set CPU tasks to yes and selected two CPU based projects but still nada???

. . Anyone know how to tame this beast ???

Stephen

??

Well as usual, anytime you change a project's preferences to something other than originally set, or add a brand new project, you run afoul of BOINC's attempt to balance credit debt to other projects.

If you have patience, it will eventually balance out the debt and will then obey your long-term cache levels. But Einstein is one of the most difficult projects to get to play well with other projects in the short-term.

In the meantime, to prevent getting totally swamped with work and forcing your other projects into the background, just set NNT on the project after retrieving a lump of cpu and gpu work. Then when that gets low again, unset NNT to allow another batch of work and set NNT again. Eventually the credit debt will get satisfied and you can permanently allow work at your 0.1/0.0 day cache levels.

If you are sure you have allow cpu work set in your preferences . . . . and specifically for the location or venue you have assigned your host at Einstein, set sched_op_debug in logging Options and check the log after you make a scheduler request to Einstein to be sure you are actually asking for cpu work seconds.

Well, the place of sacrifice has re-opened in the last week, and I went down and had them sacrifice a chicken on my behalf - and a couple of libations on the side. Seems not to have worked, but at least I had fun trying.

What, that one up at the moor? Because else it's probably the lame chickens. The one up there has classy chickens. :)
And a very delish chocolate brownie pudding.

Some more hair needs to be burnt and howling at the moon never hurts.

Well, have you asked Jason Priestley? (Or Ian Ziering, but he may be JP now).
Yeah I watched The Order

Down just when Mary's book came out. It is a conspiracy I tell you!

Well, have you asked Jason Priestley? (Or Ian Ziering, but he may be JP now).
Yeah I watched The Order

Ha ha ha . . . . yeah I got that. Be careful what you wish for and all . . . . .

It went over my bald head.

Well as usual, anytime you change a project's preferences to something other than originally set, or add a brand new project, you run afoul of BOINC's attempt to balance credit debt to other projects.

If you are sure you have allow cpu work set in your preferences . . . . and specifically for the location or venue you have assigned your host at Einstein, set sched_op_debug in logging Options and check the log after you make a scheduler request to Einstein to be sure you are actually asking for cpu work seconds.

OK, the thot plickens ...

Looking back the work request response has always been 'requesting work for CPU' (I hadn't noticed that) because it then says 'sending x new tasks' but they are GPU tasks NOT CPU tasks. For some reason it is sending GPU tasks in lieu of CPU tasks. To be clear, it is NOT requesting GPU work (I did turn on debug) but is requesting only CPU work because I have none, yet the scheduler is sending ONLY GPU work.

Anyone know how to give the E@H scheduler a kick in the butt? :(

Stephen

:(

It went over my bald head.

and mine too ...
all I know is that Jason Priestly is an actor ... I think ...

Stephen

<shrug>

Well as usual, anytime you change a project's preferences to something other than originally set, or add a brand new project, you run afoul of BOINC's attempt to balance credit debt to other projects.

If you are sure you have allow cpu work set in your preferences . . . . and specifically for the location or venue you have assigned your host at Einstein, set sched_op_debug in logging Options and check the log after you make a scheduler request to Einstein to be sure you are actually asking for cpu work seconds.

OK, the thot plickens ...

Looking back the work request response has always been 'requesting work for CPU' (I hadn't noticed that) because it then says 'sending x new tasks' but they are GPU tasks NOT CPU tasks. For some reason it is sending GPU tasks in lieu of CPU tasks. To be clear, it is NOT requesting GPU work (I did turn on debug) but is requesting only CPU work because I have none, yet the scheduler is sending ONLY GPU work.

Anyone know how to give the E@H scheduler a kick in the butt? :(

Stephen

:(

You could set the pref to no GPU work and see if you get anything for the CPU.

It has become a classic outrage

OK, the thot plickens ...

Anyone know how to give the E@H scheduler a kick in the butt? :(

Stephen

:(

You could set the pref to no GPU work and see if you get anything for the CPU.

. . I am wary of BOINC, is there a chance that would cause it to trash the hundreds of GPU tasks that are presently on the machines?

Stephen

?

OK, the thot plickens ...

Anyone know how to give the E@H scheduler a kick in the butt? :(

Stephen

:(

You could set the pref to no GPU work and see if you get anything for the CPU.

. . I am wary of BOINC, is there a chance that would cause it to trash the hundreds of GPU tasks that are presently on the machines?

Stephen

?

I mean set it on Einstein site. Should only effect new requests.

It has become a classic outrage

It is well into the 3rd day, yes truly a classic.

It could also be that the University went on summer vacation and they just pulled the plug on all non-essential hardware.

It could also be that the University went on summer vacation and they just pulled the plug on all non-essential hardware.

Then they shopuld have put a message to that effect on the front page, to replace the "The SETI@home message boards will continue to operate" statement that's currently visible.

(as it wouldn't be if the server had been turned off)

SETI@home is temporarily shut down for maintenance.

HA HA HA

It could also be that the University went on summer vacation and they just pulled the plug on all non-essential hardware.

Then they shopuld have put a message to that effect on the front page, to replace the "The SETI@home message boards will continue to operate" statement that's currently visible.
(as it wouldn't be if the server had been turned off)

This server is less important than Seti lol

It could also be that the University went on summer vacation and they just pulled the plug on all non-essential hardware.

1) Maybe they forget to turn the project back on after the maintenance?

2) They have hardware / software issues?